Member
Joined 2003
Hello,
I am nearing the completion of my next speaker project, a tower design consisting of the Peerless 8" HDS (830884) and the Peerless HDS tweeter (810921). Since I lack the tools and experience to make my own measurements, I am relying solely on manufacturer data and the measurements of third parties. The resulting frequency response and phase response is a simulated in box response produced using the excellent FRD tools. The last step of this design is of course, the crossover.
What I've come up with for a crossover is a acoustic 4th order linkwitz-riley at 1600Hz. The drivers are going to be on a flat baffle, with the tweeter inset into the baffle, so I estimated that the tweeters acoustic center is about 35mm in front of the driver.
Here's where the trouble sets in. I have come up with a very good response using a small amount of parts, however the phase response doesn't line up unless I assume the drivers are in the same place. So what can be done to adjust the phase response without affecting the frequency response?
I've attached the relevant data if anyone wants to look at my crossover in Passive Crossover Designer. Just enter 0.035 in the "Driver Z offset" box to see how the response changes when taking driver offset into account. However if you don't have experience with this spreadsheet, I still think we can discuss what can be done about changing phase response without affecting frequency response. I'll post the project in full detail once I've decided that the design is complete.
Thanks!
I am nearing the completion of my next speaker project, a tower design consisting of the Peerless 8" HDS (830884) and the Peerless HDS tweeter (810921). Since I lack the tools and experience to make my own measurements, I am relying solely on manufacturer data and the measurements of third parties. The resulting frequency response and phase response is a simulated in box response produced using the excellent FRD tools. The last step of this design is of course, the crossover.
What I've come up with for a crossover is a acoustic 4th order linkwitz-riley at 1600Hz. The drivers are going to be on a flat baffle, with the tweeter inset into the baffle, so I estimated that the tweeters acoustic center is about 35mm in front of the driver.
Here's where the trouble sets in. I have come up with a very good response using a small amount of parts, however the phase response doesn't line up unless I assume the drivers are in the same place. So what can be done to adjust the phase response without affecting the frequency response?
I've attached the relevant data if anyone wants to look at my crossover in Passive Crossover Designer. Just enter 0.035 in the "Driver Z offset" box to see how the response changes when taking driver offset into account. However if you don't have experience with this spreadsheet, I still think we can discuss what can be done about changing phase response without affecting frequency response. I'll post the project in full detail once I've decided that the design is complete.
Thanks!
I mostly don't use Excel anymore (Open Office is free and why jump on the Microsoft upgrade train) so I can't answer your question. I don't see Jeff B here very often, but he does frequent the parts express techtalk forum. Post mentioning PCD and there are a lot of users there that can help.
You can make allpass delay circuits, but I think they are usually a waste of components. Sometimes you can use a higher order crossover on the tweeter to delay it a bit. Best might be to use an optimizer like Speaker Workshop instead of the PCD - although the effort is larger, it is nice to be able to draw an arbitrary circuit.
You can make allpass delay circuits, but I think they are usually a waste of components. Sometimes you can use a higher order crossover on the tweeter to delay it a bit. Best might be to use an optimizer like Speaker Workshop instead of the PCD - although the effort is larger, it is nice to be able to draw an arbitrary circuit.
If the woofer and tweeter's rolloffs are symmetric LR4, you can't make phase aligned. That's why people use assymmetric rolloffs to achieve phase alignment on a flat baffle.
Can you email me your frd and zma files to me? Not raw files but the ones with baffle and box effects added. Then I'll show you how you can do this. It will take only a few components to obtain a good FR and phase tracking for these drivers.
My email is at my website.
Edit: I found you already attached them. I'll get back to you soon.
Can you email me your frd and zma files to me? Not raw files but the ones with baffle and box effects added. Then I'll show you how you can do this. It will take only a few components to obtain a good FR and phase tracking for these drivers.
My email is at my website.
Edit: I found you already attached them. I'll get back to you soon.
Member
Joined 2003
Thanks for your replies. Sorry I am still learning crossover design, and was not aware that the asymmetrical roll off was an accepted solution to achieve better phase alignment.
The frequency response plots include the phase information, I wasn't aware that I needed phase information to be included in the impedance plot as well to have accurate results. I'll have to look up how to do this tomorrow, and play with the crossover some more.
I had started my crossover design with Speaker Workshop, but I really dislike the interface. I found it difficult to quickly see the effects of a change in parts. This is where Passive Crossover Designer came in much use.
The frequency response plots include the phase information, I wasn't aware that I needed phase information to be included in the impedance plot as well to have accurate results. I'll have to look up how to do this tomorrow, and play with the crossover some more.
I had started my crossover design with Speaker Workshop, but I really dislike the interface. I found it difficult to quickly see the effects of a change in parts. This is where Passive Crossover Designer came in much use.
This is what I came up with quickly. Certainly not the best I can do. But I hope it can serve you as some starting point.
Don't use full 6 dB baffle step compensation. 4 to 5 dB BSC will give better midrange clarity. Anyway, you can start with 6 dB BSC and can later unwind the primary coil to reduce it when voicing the speaker.
Don't use full 6 dB baffle step compensation. 4 to 5 dB BSC will give better midrange clarity. Anyway, you can start with 6 dB BSC and can later unwind the primary coil to reduce it when voicing the speaker.
Member
Joined 2003
Thanks Jay_WJ, you're assistance is greatly appreciated. From your work I have concluded that if I leave the tweeter crossover as it is, but move the crossover on the woofer up to 1800Hz the phase response improves greatly and the resulting summed frequency response is also still very good. However, ideally I would like to keep the woofer crossover at 1600Hz to keep the cone breakup as low as possible (6dB lower in this case), so I tried as well to keep the woofer crossover where it was and move the tweeter crossover point. Moving the tweeter crossover up to 1800Hz also gave excellent phase characteristics, but there was not a dip in the frequency response so it could not be used.
Playing with the values some more I think I have come up with a very good combination. The response is normalized around 85dB, which is 4dB down from the woofers 89dB sensitivity, so that's how I came up with the amount of Baffle Step Compensation.
I think I will be very pleased with the results of this speaker system. Next step is order the parts and make the CAD drawing for my cabinet builder. All the parts will come from Solen, except the big 3mH inductors will have to be iron core so I'll have to import them from Madisound since Solen only carries air core inductors.
I've attached my final crossover result using the part values and DC resistances of what is obtainable for me.
I can't wait to build these speakers
Playing with the values some more I think I have come up with a very good combination. The response is normalized around 85dB, which is 4dB down from the woofers 89dB sensitivity, so that's how I came up with the amount of Baffle Step Compensation.
I think I will be very pleased with the results of this speaker system. Next step is order the parts and make the CAD drawing for my cabinet builder. All the parts will come from Solen, except the big 3mH inductors will have to be iron core so I'll have to import them from Madisound since Solen only carries air core inductors.
I've attached my final crossover result using the part values and DC resistances of what is obtainable for me.
I can't wait to build these speakers
I just checked your xo sim. I found you set driver geometry parameters incorrectly. You need to set the (vertical and Z) offsets of ONly the woofer, NOt those of tweeter since the simulated mic is at the tweeter axis. Leave tweeter offsets all zero.
This means you'll need to redesign your xo.
This means you'll need to redesign your xo.
Member
Joined 2003
Crap, I was editing you're file and I saw you didn't enter offsets for the tweeter, so I entered them in but didn't notice you put them in for the woofer offset. Back to the drawing board 
If what you say is correct, then shouldn't the offsets entered for the woofer be negative values? Since the woofer will be below the tweeter and the acoustic center will be behind the tweeter's?
If what you say is correct, then shouldn't the offsets entered for the woofer be negative values? Since the woofer will be below the tweeter and the acoustic center will be behind the tweeter's?
This is not my project. But what I know and my experience with systems suggests that something is wrong here. After you build it, test it and look at the impulse response. If you go with a fourth order slope at either 1600 or 1800 Hz I believe you will find that the tweeter is still leading the highest frequency from the woofer by a considerable amount (unless you measure from way off-axis).
The key variables in a lagging acoustic center for a woofer is the high end cut-off frequency and the slope. The lower the frequency and the higher the slope the more apparent lag in acoustic center for the frequencies in the crossover region.
Mark
The key variables in a lagging acoustic center for a woofer is the high end cut-off frequency and the slope. The lower the frequency and the higher the slope the more apparent lag in acoustic center for the frequencies in the crossover region.
Mark
MarkMcK said:
Your description is correct. That's why we want to find optimal asymmetric rolloffs and xo point for the drivers, given a physical limitation (i.e. acoustic center offest on a flat baffle). This way, phase alignment cannot be perfect. But when using asymmetric LR4-ish slopes, phase aligned within, say, 50 degs, one octave above and below the chosen xo freq should be regarded as excelllent. This is how people handle the acoustic center offset problem with a flat baffle.
-jAy
Please consdier that some "people" may handle it this way. Some "people" may consider plus or minus 50 degrees as excellent. Other "people" may not.
Once everyone is in agreement about the simulations, I would still suggest impulse testing of the built loudspeaker. Unless the sims start to indicate a needed offset equivalent of much more than 35 mm for a a crossover between 1600 and 1800 Hz and that is low pass filtering the woofer at at least fourth order acoustical, then I believe that the tweeter will still be leading the woofer by a considerable amount.
I will try to keep up with this thread to see what the final offset figure will be.
Mark
MarkMcK said:
There's no other way in handling this problem than using asymmetric rolloffs unless you want to use a slanted/stepped baffle or a delay network.
For example, see Zaph's asymmetric LR4 designs (BAMTM and XG18 MTM) that use a xo point below 2 kHz. Also, take a look at my two-way designs. These all work in the same way. I think they do have good phase alignment.
But I agree that an 8" woofer makes the situation a bit worse. Its AC offset is a little too high. Also, its beaming may be a problem. But I believe we can still obtain reasonably good phase alignment around the xo point.
For an extreme example, look at Mark K's Dayton RS225 8" 2-way design:
http://www.markk.claub.net/RS225_RS28A_updates/dayton_reference_rs225.htm
He uses a 1.2 kHz XO point with a very steep slope! No wonder this design's optimal listening axis is 10 deg below the tweeter axis. Indeed, its vertical offaxis performance is not very good due to not so good phase alignment. But people who built this design are very happy with its sound.
And using delays is the other way to compensate for offset with flat baffles.
Also, there are standards of evidence. The testimony of people who have built a system is not a high standard of evidence.
I realize this is difficult.
But the question of my interest is whether the simulation was done correctly? The secondary question will be if the well done simulation comes close to predicting actual acoustic center or location of sound origin through the crossover region?
Also, there are standards of evidence. The testimony of people who have built a system is not a high standard of evidence.
I realize this is difficult.
But the question of my interest is whether the simulation was done correctly? The secondary question will be if the well done simulation comes close to predicting actual acoustic center or location of sound origin through the crossover region?
Of course, it is not a good standard. But I know that Mark is a person who knows what he is doing.
Simulation is reasonably accurate, IMO. Visit my website and see some of my comparisions of sim and real measurement in the Appendix of my xo design notes. These are only a part of sims I did to replicate many existing designs. I was convinced that if done correctly, sims are reasonably accurate in predicting frequency and phase responses.
Member
Joined 2003
Ok, sounds like my offsets in PCD were incorrect to begin with, as I was using a positive Z offset for the tweeter.
Since PCD calculates the response with the tweeter as the center listening axis, what benefit do we have from being able to enter offsets for the tweeter? I can't think off the top of my head any reason why we would want to do this, so it only makes for more confusion letting the user have the ability to make errors by assuming the woofer will be the center listening axis.
I've found a bit of information on passive delay circuits, so I think I may investigate them as well if I can't seem to find a good combination of parts without them that will work. I'll have to simulate them in SpeakerWorkshop, as they can't be simulated in PCD as far as I can tell.
Of course I would like to measure the speaker when it's finished, but I don't have the equipment of facilities to do so, so I am trying to do the most accurate simulation with the tools available.
Since PCD calculates the response with the tweeter as the center listening axis, what benefit do we have from being able to enter offsets for the tweeter? I can't think off the top of my head any reason why we would want to do this, so it only makes for more confusion letting the user have the ability to make errors by assuming the woofer will be the center listening axis.
I've found a bit of information on passive delay circuits, so I think I may investigate them as well if I can't seem to find a good combination of parts without them that will work. I'll have to simulate them in SpeakerWorkshop, as they can't be simulated in PCD as far as I can tell.
Of course I would like to measure the speaker when it's finished, but I don't have the equipment of facilities to do so, so I am trying to do the most accurate simulation with the tools available.
A delay network is usually not used with a LR4 design because asymmetric rolloffs can handle phase tracking. Take a look at Zaph's ZD5 design page for more information about delay network.
I think with an 8" driver on a flat baffle in a 2-way should require the design axis to be below the tweeter axis. Slight downward listening is not bad at all.
BTW, did you get my email?
I also found that your 830884 impedance is slightly higher than Zaph's measurement.
I think with an 8" driver on a flat baffle in a 2-way should require the design axis to be below the tweeter axis. Slight downward listening is not bad at all.
BTW, did you get my email?
I also found that your 830884 impedance is slightly higher than Zaph's measurement.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.